在 LTX-Video 文生视频提示词中，先写主体和动作，再写镜头运动与景别，最后补光线和风格。把“slow dolly in”这类可执行动作与“cinematic”这类风格词分开，便于定位失败原因。 For LTX-Video text-to-video prompts, describe subject and action first, camera movement and shot size second, then lighting and style. Keep executable motion such as “slow dolly in” separate from broad style terms like “cinematic”.